Kousar Malekinasab
-
Dr. Jianping Pan, Department of Computer Science, University of Victoria (Supervisor)
-
Dr. Sudhakar Ganti, Department of Computer Science, UVic (Member)
Topic
Characterizing and Optimizing Real-Time Communication over Low Earth Orbit Satellite Networks: A Starlink Measurement and Design Study
Department of Computer Science
Date & location
-
Wednesday, August 12, 2026
-
10:00 A.M.
-
Virtual Defence
Reviewers
Supervisory Committee
-
Dr. Jianping Pan, Department of Computer Science, University of Victoria (Supervisor)
-
Dr. Sudhakar Ganti, Department of Computer Science, UVic (Member)
External Examiner
-
Dr. Navneet Popli, Department of Electrical and Computer Engineering, University of Victoria
Chair of Oral Examination
-
Dr. Mario Bras, Department of Mechanical Engineering, UVic
Abstract
Real-Time Communication (RTC) services such as Google Meet, Zoom, and Microsoft Teams are primarily designed around terrestrial Internet assumptions. Low Earth Orbit (LEO) satellite networks, such as Starlink, introduce a fundamentally different access model in which traffic traverses a satellite segment before exiting through a terrestrial Point of Presence (PoP). Consequently, RTC performance depends heavily on Starlink PoP associations, satellite-network dynamics, and provider-side service-point selection policies. This thesis characterizes and optimizes RTC over Starlink. First, it presents a measurement study of major RTC platforms using stationary and in-flight experiments. The results reveal divergent service-point selection behaviors: Google Meet generally tracks Starlink PoPs, Microsoft Teams typically employs a rigid session-anchored strategy, and Zoom utilizes regional hubs. Second, through a geo-distributed glass-to-glass delay analysis, this thesis demonstrates that while the Starlink access segment provides relatively low and stable delay, terrestrial post-PoP routing frequently dominates the application-visible latency when traffic is anchored to distant service points. These findings establish that existing RTC systems are not consistently LEO- or peer-aware. Motivated by these observations, this thesis proposes LeoRTC, a LEO-aware multi-relay framework that dynamically selects relay PoPs and leverages space-based paths to minimize end-to-end media latency. Trace-driven simulations and emulations confirm that LeoRTC substantially reduces one-way latency while maintaining low jitter and negliible packet loss. Overall, this work demonstrates that joint consideration of satellite access and post-PoP routing is essential for designing next-generation RTC systems.